• Proceedings of the KSPS conference
• /
• 2006.05a
• /
• pp.139-142
• /
• 2006

Speaker verification systems can be implemented using speaker adaptation methods if the amount of speech available for each target speaker is too small to train the speaker model. This paper shows experimental results using well-known adaptation methods, namely Maximum A Posteriori (MAP) and Maximum Likelihood Linear Regression (MLLR). Experimental results using Korean speech show that MLLR is more effective than MAP for short enrollment utterances.

• Smart Media Journal
• /
• v.2 no.2
• /
• pp.33-38
• /
• 2013

In this paper, we investigated baseline estimation methods for remove background noise using Raman spectra from acute alcohol liver injury and acute ethanol-induced chronic liver fibrosis. Far the baseline estimation, we applied first derivative, linear programming and rolling ball method. Optimal input parameter of each method were determined by the training rate of MAP (maximum a posteriori probability) classifier. According to the experimental results, classification results baseline estimation with the rolling ball algorithm gave about 89.4%, which is very promising results for classification of acute alcohol liver injury and acute ethanol-induced chronic liver fibrosis. From these results, to determined the appropriate methods and parameters of baseline estimation impact on classification performance was confirmed.

• The Journal of the Acoustical Society of Korea
• /
• v.30 no.6
• /
• pp.324-329
• /
• 2011

In this paper, we propose a novel approach to improve the performance of a statistical model-based voice activity detection (VAD) which is based on the conditional maximum a posteriori (CMAP) with deviation. In our approach, the VAD decision rule is expressed as the geometric mean of likelihood ratios (LRs) based on adapted threshold according to the speech presence probability conditioned on both the speech activity decisions and spectral deviation in the pervious frame. Experimental results show that the proposed approach yields better results compared to the CMAP-based VAD using the LR test.

• Journal of IKEEE
• /
• v.4 no.2 s.7
• /
• pp.257-265
• /
• 2000

Turbo Code decoder is an iterate decoding technology, which extracts extrinsic information from the bit to be decoded by calculating both forward and backward metrics in each decoding step, and uses the information to the next decoding step. Viterbi decoder, which is for a convolutional code, runs continuous mode, while Turbo Code decoder runs by block unit. There are algorithms used in a decoder : which are MAP(maximum a posteriori) algorithm requiring very complicated calculation and SOVA(soft output Viterbi algorithm) using Viterbi algorithm suggested by Hagenauer, and it is known that the decoding performance of MAP is better. The result of this make experimentation shows that the performance of SOVA, which has half complex algorithm compare to MAP, is almost same as the performance of MAP when the SOVA decoding performance is supplemented with Robust equalization techniques.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.44 no.4 s.316
• /
• pp.117-122
• /
• 2007

For multiple antenna systems, we consider the hybrid iterative detection of the maximum a posteriori probability(MAP) detection and the linear detection such as the minimum-mean-square-error(MMSE) filtering with soft cancelation. We devise methods to obtain both the lower complexity of the linear detection and the superior performance of the MAP detection. Using the a prior probability of the coded bit which is extrinsic of the outer decoder, we compute the threshold of grouping and determine the detection scheme symbol by symbol. Through the simulation results, it is shown that the proposed receiver obtains the superior performance to the MMSE detector and the lower complexity than the MAP detector.

• Proceedings of the IEEK Conference
• /
• 2000.11b
• /
• pp.277-280
• /
• 2000

In this paper, we designed a turbo decoder using VHDL. To maximize effective free distance of the turbo code, we implemented pseudo random interleaver. A MAP(Maximum a posteriori) decoder is used as a primimary decoder. We avoided multiplication by using lookup tables(ROM). We expect that this small-sized turbo decoder is suitable for mobile communication. We simulated turbo decoder with Altera MAX+PLUS II.

• ETRI Journal
• /
• v.26 no.5
• /
• pp.384-390
• /
• 2004

We propose a new bandwidth-efficient technique that achieves high data rates over a wideband wireless channel. This new scheme is targeted for a multiple-input multiple- output orthogonal frequency-division multiplexing (MIMO-OFDM) system that achieves transmit diversity through a space frequency block code and capacity enhancement through the iterative joint processing of zero-forcing detection and maximum a posteriori (MAP) decoding. Furthermore, the proposed scheme is compared to the coded Bell Labs Layered Space-Time OFDM (BLAST-OFDM) scheme.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.9
• /
• pp.4587-4605
• /
• 2019

High Efficiency Video Coding (HEVC) suffers from high computational complexity due to its quad-tree structure in motion estimation (ME). This paper exposes an adaptive search range decision algorithm for accelerating HEVC integer-pel ME on GPU which estimates the optimal search range (SR) using a MAP (Maximum A Posteriori) estimator. There are three main contributions; First, we define the motion feature as the standard deviation of motion vector difference values in a CTU. Second, a MAP estimator is proposed, which theoretically estimates the motion feature of the current CTU using the motion feature of a temporally adjacent CTU and its SR without any data dependency. Thus, the SR for the current CTU is parallelly determined. Finally, the values of the prior distribution and the likelihood for each discretized motion feature are computed in advance and stored at a look-up table to further save the computational complexity. Experimental results show in conventional HEVC test sequences that the proposed algorithm can achieves high average time reductions without any subjective quality loss as well as with little BD-bitrate increase.

• Progress in Medical Physics
• /
• v.15 no.2
• /
• pp.70-76
• /
• 2004

The most important work for early detection of liver cancer and decision of its characteristic and location is good segmentation of a liver region from other abdominal organs. This paper proposes a fully automatic liver segmentation algorithm based on the abdominal morphology characteristic as an easy and efficient method. Multi-modal threshold as pre-processing is peformed and a spine is segmented for finding morphological coordinates of an abdomen. Then the liver region is extracted using C-class maximum a posteriori (MAP) decision and morphological filtering. In order to estimate results of the automatic segmented liver region, area error rate (AER) and correlation coefficients of rotational binary region projection matching (RBRPM) are utilized. Experimental results showed automatic liver segmentation obtained by the proposed algorithm provided strong similarity to manual liver segmentation.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.12
• /
• pp.108-115
• /
• 1994

Most pattern classifiers have been designed based on the ML (Maximum Likelihood) training algorithm which is simple and relatively powerful. The ML training is an efficient algorithm to individually estimate the model parameters of each class under the assumption that all class models in a classifier are statistically independent. That assumption, however, is not valid in many real situations, which degrades the performance of the classifier. In this paper, we propose a minimum-error-rate training algorithm based on the MAP (Maximum a Posteriori) approach. The algorithm regards the normalized outputs of the classifier as estimates of the a posteriori probability, and tries to maximize those estimates. According to Bayes decision theory, the proposed algorithm satisfies the condition of minimum-error-rate classificatin. We apply this algorithm to NPM (Neural Prediction Model) for speech recognition, and derive new disrminative training algorithms. Experimental results on ten Korean digits recognition have shown the reduction of 37.5% of the number of recognition errors.